For instance, the architecture could be updated to support ingestion filtering based on labeler data:

I'm honestly curious how this architecture will support various legal and compliance requirements from around the world. I get how the PDSes can claim to hosted in some specific locality, but the BGSes and beyond are going to be global services, which process and store data.

For instance, the moderation plans publicized are based off of the data provided by the labeling service, but labeling occurs after the BGSes ingests content from the PDSes, but filters could be applied before accepting content, at ingestion time.

Sure, you are proving you're good at logistics, but you're also doing people harm and ruining your own reputation in the process.

Everyone was frustrated, but just for different reasons.

He's one of a small group of people providing funding and on the board so it depends on what you mean by "in charge".

- For free, we'll block the top 10 bad words, if you pay we'll proactively filter for bad content.

Things like that become possible when you separate content hosting, content tagging, content filtering, and feed generation into separate services.

Imagine:
- Free user hosting inserts ads into what you post, paid hosting doesn't.
- Free feed aggregation inserts ads into what you see, paid doesn't.
- For free you can find your own feeds to watch, but if you pay you can get a kick ass personal algorithm giving you better content.

I think the play is market disruption. Older social media puts power in the hands of single corporations and limits competition. Their vision of decentralization supports more diverse ecossystems, and if they set up the ecosystem they have first-in advantage and better insight into competion.

So, do you want to be locked into one of ten fediverse instances, which I hear are not doing great jobs, or do you want to hope someone provides a good services for Bluesky at some point?

But that doesn't solve the underlying problem, it just gives Bluesky an excuse for saying someone else should solve that part of the problems.

Bluesky is working on making each part of the social network it's own mini-service. So you can host your own content in one place, pull labelling data from another service, and then enforce moderation based on a third service. (or get them all from one place, like bsky.app if you want.)

I'm not an expert, but I believe Mastodon federates its services differently. (But probably not better.)

Mastodon and the Fediverse are a collection of "mini-twitters" that trade content but act independently. Each service maintains it's own moderation tools that only work inside their service.

Once federation is available the content will be distributed over the atprotocol Bluesky is developing, but not hosted or allowed by Bluesky itself. That coupled with the asymmetric cost of posting vs filtering content makes me extremely worried about the future viability of the platform.

People are really going to hate it when Bluesky's documented plan for safety is actually implemented. It basically boils down to 1) let bad people post what they want, 2) *try* to label content, and 3) *hope* people figure out how to block the worst of it by filtering labels.

To clarify, my basic questions is What The Fuck?